AI 'godfather' Hinton: humans might have been better off without AI

Geoffrey Hinton, a pioneer of neural-network research, argues that AI could bring great benefits but may ultimately prove more harmful than useful. He believes the next decade will show whether society can manage its risks and achieve widespread prosperity, or instead face mass unemployment, widening inequality and social upheaval. He warns that a serious incident could happen as soon as this year or next: an AI-enabled attack might cause a major power outage or disable financial systems for a week. He compares recent AI agents’ attempts to hack a website to an early warning such as the Three Mile Island accident, and says a crisis could force governments to take the danger seriously.

Hinton criticizes leading technology executives for releasing increasingly powerful systems in pursuit of market dominance, and says he does not trust OpenAI chief Sam Altman. He supports mandatory testing of frontier AI, independent assessments and international standards, rather than companies regulating themselves. His proposed model is similar to drug approval: companies should have to show an independent authority that a system’s benefits outweigh its harms before release. He warns that waiting for damage and relying on lawsuits would repeat mistakes such as the thalidomide disaster. Despite his alarm, Hinton remains cautiously hopeful that people can retain control, provided governments and society make a major effort to regulate the technology.